Neon Pricing Calculator

How Neon pricing works in 2026

Neon is a serverless Postgres database with two distinct billing dimensions: compute and storage. Compute is measured in Compute Units (CU) — one CU equals 1 vCPU and 4 GB RAM — and billed per hour of active use at $0.106/CU-hour on the Launch plan. Storage is $0.35/GB-month. Egress is 100 GB/month included, then $0.10/GB.

The defining feature of Neon is scale-to-zero: your database automatically suspends after a configurable idle period (default 5 minutes). While suspended, you pay zero compute cost — only storage is billed. This makes Neon dramatically cheaper than always-on databases for workloads with bursty or low traffic patterns.

Neon compute units explained

Neon's minimum compute size is 0.25 CU (0.25 vCPU, 1 GB RAM). From there it scales to 0.5 CU (2 GB), 1 CU (4 GB), 2 CU (8 GB), and up to 16 CU (64 GB). Autoscaling adjusts the CU size dynamically — you set a minimum and maximum, and Neon scales within that range based on load. The Launch plan allows up to 8 CU autoscale.

Billing is based on the CU size at each moment, not a fixed tier. If your database only spikes to 2 CU for 2 hours out of the month, you pay for 2 CU × 2 hours, not 2 CU × 720 hours.

Neon pricing examples

Side project (scale-to-zero, 1 GB RAM, 50 active hours): 0.25 CU × 50 hours × $0.106 + 0.5 GB storage × $0.35 = $1.50/month.

Startup SaaS (always-on, 1 vCPU / 4 GB RAM): 1 CU × 720 hours × $0.106 + 5 GB storage × $0.35 = $78.07/month.

Growing SaaS (2 CU, scale-to-zero, 300 active hours): 2 CU × 300 hours × $0.106 + 20 GB storage × $0.35 = $70.60/month.

What to watch out for with Neon

The main limitation of the Launch plan is no HA. Data is durably replicated, but there is no user-selectable multi-AZ primary. If your primary compute fails, Neon handles recovery automatically, but you cannot configure an active standby. For strict HA requirements, consider PlanetScale HA clusters or Supabase Pro.

Always-on workloads with large CU sizes can become expensive. 4 CU × 720 hours × $0.106 = $305/month in compute alone before storage. At that scale, compare against dedicated-instance options like Supabase or PlanetScale.

How this calculator works

The calculator maps your target RAM to the nearest Neon CU size, then multiplies by your active compute hours per month. If scale-to-zero is enabled, your active hours represent only the hours the database is actually receiving queries. Storage and egress overage are added separately.

Frequently asked questions

How much does Neon cost per month?

A side project with scale-to-zero can cost as little as $1–3/month. A production SaaS on 1 CU always-on runs about $78/month before storage. Neon has no monthly minimum on the Launch plan — you pay only for what you use.

Does Neon have a free plan?

Yes. The Free plan includes 100 CU-hours/month, 0.5 GB storage, and up to 2 CU autoscale. It is suitable for development and side projects. The Launch plan removes these limits and enables scale-to-zero with configurable autoscale up to 8 CU.

Neon vs Supabase — which is cheaper?

Neon is usually cheaper for pure database workloads, especially with scale-to-zero. Supabase Pro ($25/month) bundles Auth, Storage, Edge Functions, and Realtime — so if you need those services, Supabase's total cost is often lower. For database-only workloads on low traffic, Neon wins. For full BaaS at medium scale, Supabase is the better value.

Is Neon Postgres-compatible?

Yes. Neon is standard Postgres — no proprietary APIs or query language. You connect with a standard Postgres connection string and use any ORM or driver that works with Postgres: Drizzle, Prisma, Sequelize, or raw SQL. Point-in-time restore, logical replication, and Postgres extensions are all supported.